Fun With 4th Downs

There's been a lot of good work on the subject of 4th down decision making; most notably, there's this paper by David Romer in which he concludes that going for it is beneficial a lot more often than coaches think it is.  I agree wholeheartedly, and I've been outspoken about being aggressive inside your opponent's 40 yard line.  I've said repeatedly that it's never a good idea to punt inside the 40, but that is admittedly too rigid a stance, and as Saturday's game illustrated, those 4th down situations can be incredibly nuanced.

Situation #1: Up 14-7 in the 1st quarter, it's 4th-and-11 for NC State at the Maryland 38.  NC State initially elects to punt and manages to pin Maryland inside its own 5.  But the Terps rough the punter and the coaches decide to take the penalty, which set up 4th-and-6 at the 33, and put the offense back on the field.  Wilson's pass is intercepted and returned for a touchdown--doomsday scenario.

Considerations: State's defense is terrible, obviously, so you'd like to be aggressive as much as reasonably possible when State's offense is in decent position to keep the ball in its hands and put points on the board.  Maryland's offense is terrible, too, and even before Chris Turner went down, they were moving the ball at an unimpressive 4.6 yards/play clip.  We're out of field goal range at the 33, as a 48-yard FG that barely cleared the crossbar later proved.  So that's completely off the table.

The disastrous outcome of the 4th down play isn't a factor either.  Just like you can't kick yourself for folding jack-ten to a big raise even though you would've ended up with the nut straight, the pick-six is irrelevant hindsight.  The key is making the right decision with the data available to you at the time.  The fact that we knew we had Maryland pinned is a factor in the process, though.

So, good decision?  Let's look at some numbers.  On average, teams that start a drive between the 31-40 yard line score twice as often as they do when they start inside their own 10.  Based on this fine work from Bill Connelly, we also know that teams convert 4th-and-6s 36% of the time on average.  With State's good offense working against Maryland's poor defense, the odds are probably a bit higher than that.

The cost of not converting here and giving Maryland the ball at the 33 instead of the 3 (or wherever it was downed) is about one point, according to Connelly's study.  If we do convert, we can reasonably expect to come away with at least three points.  Two-to-one odds against converting (probably closer to 3:2), two-to-one odds against Maryland scoring points if they get the ball at the 33.  Five-to-one against scoring if Maryland gets the ball inside the 10, and that's the wrinkle that makes this tough.  I've gone back and forth so much that I don't think there's a clear cut answer, but I lean towards thinking that going for it was the right move. 

Situation #2: Up 38-31 with a little over a minute left, NC State has the ball at Maryland's 31 yard line; it's 4th-and-12. Maryland is out of time outs, so a conversion makes no defensive stand necessary.  NC State decides to go and Russell Wilson fires a pass into the end zone; Owen Spencer gets his hands on it but can't hang on.  Turnover on downs.

Considerations: Obviously the coaches weren't worried about handing the ball over on the 31, and Czajkowski proved earlier he could hit a field goal from this spot, so why not give that a shot?  The doomsday scenario would be a blocked kick, which I wouldn't put past us.  And Czajkowski might not get all of the kick this time and come up short.  The odds of the offense converting that 4th-and-long were probably in the neighborhood of 4.5-to-1 against.  I tend to think the odds of a made field goal were better.

At that late stage of the game, especially with the Terps out of time outs, I think either you try the field goal or you go ahead and take your chances with a punt.  Yes, I'm suggesting a Russell Wilson quick kick might've been a good idea.  Even if the ball sails into the end zone.  Going back to the averages, the odds that a drive starting at the 20 results in a TD are 4-to-1 against.  With just a minute to work with and zero time outs, they're probably much longer than that.